- From: Alberto Reggiori <alberto@asemantics.com>
- Date: Tue, 16 Mar 2004 02:10:03 +0100
- To: RDF Data Access Working Group <public-rdf-dawg@w3.org>
== Use Case Name News flavoring and alert system == Intent: Task & Roles Actor/User Agent wants to be notified/alerted about certain events or news triggers accordingly to his/her personal profile preferences - or he/she wants to seamlessly search/browse a specific service which automatically filter news/information accordingly to well-known kinds/flavors. == Key Benefits / Value Users are used to daily poll certain repositories or sources of information online to search for and read about specific news or get a gist of the daily events/threads - such a task is generally time consuming and not trivial especially if the number of sources to monitor/read is high. A more automated service is needed. By simply expressing his/her own preferences in a profile file available at certain URI, the user can delegate such tasks to an ad-hoc designed piece of software which will take care of collecting specific online news/events accordingly to his/her profile. Or by subscribing to a specialized alert service using the profile information, he/she can automatically be notified of specific events. Both the user profile and the target sources of information/news must be fully decoupled and decentralized from each other. And the user must retain full control on his/her personal profile information; as well as the sources. == Description Today there are several different News sources on the Web which provide near real-time fresh information about world-wide events, online newspapers or simple community-specific news; Google News [1] is a good example of a near real-time Web service providing news articles and related images/pictures covering most of the world daily events. Similarly using ad-hoc blogging systems, newsletters or some kind of digested messages, most IT and Web communities daily exchange information. There exist several systems today which provide an integrated service to thematic news/information syndication such as Feedster [2], SlashDot [3] or casted to specific communities e.g. PlanetRDF [4] - they generally require the user to periodically "poll" the service and search for specific topics or articles of interest (or even worse, if the number of articles to read/digest is too high the user might not be able to). Even if such systems provide some sort of alert/notifcation system on thematic news/articles or some ad-hoc structured thematic browseable/searchable view over indexed information; the user must pre-register or provide to the third party his/her person information/preferences to allow the software to filter the information accordingly. Each change of preferences (or flavoring) require the user to update his/her profile onto a specific server, on a specific machine, requiring to remember his/her username/password. And the third party system to act accordingly. What is needed is a federated service/system which is able to query/aggregate on-the-fly information accordingly to the users' preferences in a fully decentralized and decoupled way (e.g. using a URL of a FOAF file [5]) - which guarantees to leave parties in control of their respective information. Either the sources and the users must be able to describe their articles/flavors using terms from any vocabulary at any level of complexity - minimal requirement is they use the same URI references. The real benefit comes when the parties use a common vocabulary e.g. simple DublinCore [6] or FOAF [7]. The query system must be able to seamlessly search and join the user's preferences and the requested information sources, and provide a unified view over the data to the end user application. Each party specific provenance/source information might be used to further refine the query details. If a common set of vocabulary terms or references can not be found the result of the query result set is empty. == Other === Examples A demonstrator of a similar system [8] has been implemented using FOAF to express users' profiles and RSS1.0 news feeds generated from a selection of italian newspapers. RDF descriptions are augmented with some application specific properties which allow to filter/flavor news accordingly. Full provenance/source information is being added to the sources and extensively used during the query process. Extended RDQL queries are used through the whole application to join on-the-fly RDF descriptions coming from the user's specified URI FOAF file together with the individual newspapers articles URIs. The UI then tailors search and browse results accordingly. === Applicability/Scale Real-time data, Personalization/FOAF == References [1] http://news.google.com/ [2] http://www.feedster.com [3] http://slashdot.org [4] http://planet.rdfhack.com/ [5] http://foaf.asemantics.com/alberto [6] http://dublincore.org [7] http://xmlns.com/foaf/0.1/ [8] http://demo.asemantics.com/biz/nb2/
Received on Monday, 15 March 2004 20:10:12 UTC